Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/tvplay.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_str
   8 from ..utils import (
   9     ExtractorError,
  10     parse_iso8601,
  11     qualities,
  12 )
  13
  14
  15 class TVPlayIE(InfoExtractor):
  16     IE_DESC = 'TV3Play and related services'
  17     _VALID_URL = r'''(?x)http://(?:www\.)?
  18         (?:tvplay\.lv/parraides|
  19            tv3play\.lt/programos|
  20            tv3play\.ee/sisu|
  21            tv3play\.se/program|
  22            tv6play\.se/program|
  23            tv8play\.se/program|
  24            tv10play\.se/program|
  25            tv3play\.no/programmer|
  26            viasat4play\.no/programmer|
  27            tv6play\.no/programmer|
  28            tv3play\.dk/programmer|
  29         )/[^/]+/(?P<id>\d+)
  30         '''
  31     _TESTS = [
  32         {
  33             'url': 'http://www.tvplay.lv/parraides/vinas-melo-labak/418113?autostart=true',
  34             'info_dict': {
  35                 'id': '418113',
  36                 'ext': 'flv',
  37                 'title': 'Kādi ir īri? - Viņas melo labāk',
  38                 'description': 'Baiba apsmej īrus, kādi tie ir un ko viņi dara.',
  39                 'duration': 25,
  40                 'timestamp': 1406097056,
  41                 'upload_date': '20140723',
  42             },
  43             'params': {
  44                 # rtmp download
  45                 'skip_download': True,
  46             },
  47         },
  48         {
  49             'url': 'http://www.tv3play.lt/programos/moterys-meluoja-geriau/409229?autostart=true',
  50             'info_dict': {
  51                 'id': '409229',
  52                 'ext': 'flv',
  53                 'title': 'Moterys meluoja geriau',
  54                 'description': 'md5:9aec0fc68e2cbc992d2a140bd41fa89e',
  55                 'duration': 1330,
  56                 'timestamp': 1403769181,
  57                 'upload_date': '20140626',
  58             },
  59             'params': {
  60                 # rtmp download
  61                 'skip_download': True,
  62             },
  63         },
  64         {
  65             'url': 'http://www.tv3play.ee/sisu/kodu-keset-linna/238551?autostart=true',
  66             'info_dict': {
  67                 'id': '238551',
  68                 'ext': 'flv',
  69                 'title': 'Kodu keset linna 398537',
  70                 'description': 'md5:7df175e3c94db9e47c0d81ffa5d68701',
  71                 'duration': 1257,
  72                 'timestamp': 1292449761,
  73                 'upload_date': '20101215',
  74             },
  75             'params': {
  76                 # rtmp download
  77                 'skip_download': True,
  78             },
  79         },
  80         {
  81             'url': 'http://www.tv3play.se/program/husraddarna/395385?autostart=true',
  82             'info_dict': {
  83                 'id': '395385',
  84                 'ext': 'flv',
  85                 'title': 'Husräddarna S02E07',
  86                 'description': 'md5:f210c6c89f42d4fc39faa551be813777',
  87                 'duration': 2574,
  88                 'timestamp': 1400596321,
  89                 'upload_date': '20140520',
  90             },
  91             'params': {
  92                 # rtmp download
  93                 'skip_download': True,
  94             },
  95         },
  96         {
  97             'url': 'http://www.tv6play.se/program/den-sista-dokusapan/266636?autostart=true',
  98             'info_dict': {
  99                 'id': '266636',
 100                 'ext': 'flv',
 101                 'title': 'Den sista dokusåpan S01E08',
 102                 'description': 'md5:295be39c872520221b933830f660b110',
 103                 'duration': 1492,
 104                 'timestamp': 1330522854,
 105                 'upload_date': '20120229',
 106             },
 107             'params': {
 108                 # rtmp download
 109                 'skip_download': True,
 110             },
 111         },
 112         {
 113             'url': 'http://www.tv8play.se/program/antikjakten/282756?autostart=true',
 114             'info_dict': {
 115                 'id': '282756',
 116                 'ext': 'flv',
 117                 'title': 'Antikjakten S01E10',
 118                 'description': 'md5:1b201169beabd97e20c5ad0ad67b13b8',
 119                 'duration': 2646,
 120                 'timestamp': 1348575868,
 121                 'upload_date': '20120925',
 122             },
 123             'params': {
 124                 # rtmp download
 125                 'skip_download': True,
 126             },
 127         },
 128         {
 129             'url': 'http://www.tv3play.no/programmer/anna-anka-soker-assistent/230898?autostart=true',
 130             'info_dict': {
 131                 'id': '230898',
 132                 'ext': 'flv',
 133                 'title': 'Anna Anka søker assistent - Ep. 8',
 134                 'description': 'md5:f80916bf5bbe1c5f760d127f8dd71474',
 135                 'duration': 2656,
 136                 'timestamp': 1277720005,
 137                 'upload_date': '20100628',
 138             },
 139             'params': {
 140                 # rtmp download
 141                 'skip_download': True,
 142             },
 143         },
 144         {
 145             'url': 'http://www.viasat4play.no/programmer/budbringerne/21873?autostart=true',
 146             'info_dict': {
 147                 'id': '21873',
 148                 'ext': 'flv',
 149                 'title': 'Budbringerne program 10',
 150                 'description': 'md5:4db78dc4ec8a85bb04fd322a3ee5092d',
 151                 'duration': 1297,
 152                 'timestamp': 1254205102,
 153                 'upload_date': '20090929',
 154             },
 155             'params': {
 156                 # rtmp download
 157                 'skip_download': True,
 158             },
 159         },
 160         {
 161             'url': 'http://www.tv6play.no/programmer/hotelinspektor-alex-polizzi/361883?autostart=true',
 162             'info_dict': {
 163                 'id': '361883',
 164                 'ext': 'flv',
 165                 'title': 'Hotelinspektør Alex Polizzi - Ep. 10',
 166                 'description': 'md5:3ecf808db9ec96c862c8ecb3a7fdaf81',
 167                 'duration': 2594,
 168                 'timestamp': 1393236292,
 169                 'upload_date': '20140224',
 170             },
 171             'params': {
 172                 # rtmp download
 173                 'skip_download': True,
 174             },
 175         },
 176     ]
 177
 178     def _real_extract(self, url):
 179         video_id = self._match_id(url)
 180
 181         video = self._download_json(
 182             'http://playapi.mtgx.tv/v1/videos/%s' % video_id, video_id, 'Downloading video JSON')
 183
 184         if video['is_geo_blocked']:
 185             raise ExtractorError(
 186                 'This content is not available in your country due to copyright reasons', expected=True)
 187
 188         streams = self._download_json(
 189             'http://playapi.mtgx.tv/v1/videos/stream/%s' % video_id, video_id, 'Downloading streams JSON')
 190
 191         quality = qualities(['hls', 'medium', 'high'])
 192         formats = []
 193         for format_id, video_url in streams['streams'].items():
 194             if not video_url or not isinstance(video_url, compat_str):
 195                 continue
 196             fmt = {
 197                 'format_id': format_id,
 198                 'preference': quality(format_id),
 199             }
 200             if video_url.startswith('rtmp'):
 201                 m = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>[^/]+))/(?P<playpath>.+)$', video_url)
 202                 if not m:
 203                     continue
 204                 fmt.update({
 205                     'ext': 'flv',
 206                     'url': m.group('url'),
 207                     'app': m.group('app'),
 208                     'play_path': m.group('playpath'),
 209                 })
 210             elif video_url.endswith('.f4m'):
 211                 formats.extend(self._extract_f4m_formats(
 212                     video_url + '?hdcore=3.5.0&plugin=aasp-3.5.0.151.81', video_id))
 213                 continue
 214             else:
 215                 fmt.update({
 216                     'url': video_url,
 217                 })
 218             formats.append(fmt)
 219
 220         self._sort_formats(formats)
 221
 222         return {
 223             'id': video_id,
 224             'title': video['title'],
 225             'description': video['description'],
 226             'duration': video['duration'],
 227             'timestamp': parse_iso8601(video['created_at']),
 228             'view_count': video['views']['total'],
 229             'age_limit': video.get('age_limit', 0),
 230             'formats': formats,
 231         }